BACKGROUND: Anticancer CAM therapies are increasingly becoming the focus of scientific research. Conventional approaches to early phase trials may not be applicable for the development of some CAM approaches. Perhaps as a result, many CAM therapies taken to clinical trial in cancer have not previously been subject to dose-finding; furthermore, Phase II trials of CAM have often failed to incorporate formal statistical designs. In recent years new designs for Phase I and II trials have been developed in response to the challenges of researching immune, biologic and target-based agents. These methodologies might be usefully adapted for the study of CAM. OBJECTIVES: To review systematically methodologies that have been applied in early phase trials of CAM for cancer; to review systematically available methodologies for such trials; to make specific practical recommendations for future CAM trials based on simulation and analytic studies. METHODS: Three separate systematic reviews will be undertaken of Phase I, II and III trials of CAM in cancer. These will describe previous research and determine the proportion of: Phase I trials that attempted to find an optimal dose; Phase II and III trials that were preceded by dose finding trials; Phase I trials that incorporated a formal statistical design; Phase Ill trials that were preceded by Phase II.A systematic review will be undertaken of Phase I and II designs: each design will be described by pre-specified criteria. The operating characteristics of key Phase I and II designs for CAM will be assessed by simulation and analytic studies. Four sets of dose-response and dose-toxicity functions will be generated for four "archetypal" CAM modalities. Each Phase I design will be tested for each modality: efficiency and closeness of the study dose to a theoretically derived optimal dose will be calculated. Phase II designs will be tested given a range of clinical scenarios common for CAM therapies: sample size and study duration will be compared between designs. The results of these studies will be expressed in non-technical terms so that CAM researchers can make a choice of Phase I and II design depending on simple assumptions about the properties of the modality and cancer population under study.